Generative AI boosters are beginning to explore how the technology could be used to augment infrastructure as code tools, according to Arun Chandrasekaran, a distinguished VP analyst at Gartner.
Speaking at the analyst firm’s Symposium in Australia today, Chandrasekaran said he’s aware of very early interest in the application of GenAI to IT infrastructure. He said we’re nowhere near developers being able to use AI to order infrastructure – “Alexa: build me a K8s cluster” – but developers are already pondering how models could use logfiles to analyze an org’s IT. As suggesting code is already one of the most prominent applications of generative AI, the potential for binary brainboxes to recommend infrastructure recipes needed to execute code is tantalizing.
AI was, unsurprisingly, the topic of the Symposium keynote – the first time Gartner has dedicated the opener at its flagship conference to a single topic. The session saw distinguished VP analyst Don Scheibenreif and senior director of research and advisory Neha Kumar articulate Gartner’s belief that generative AI will quickly make an impact on back office tasks such as assisting developers to write code, or helping users of personal productivity tools to work faster and more efficiently.
But the two analysts said back office tools like GitHub Copilot or the generative AI augmentations for Google Workspaces will not deliver competitive advantage. Nor do text-to-text tools like ChatGPT. Everyone can access them and learn how to use them, so they represent “table stakes.”
More complex AI can have greater impact.
Scheibenreif used the example of Khan Academy’s “Khanmingo” chatbot – which offers students an interactive tutor – as an example of what Gartner considers “game-changing AI.” But he and Kumar warned that building that sort of tool is harder, more expensive, and riskier than using back office AI.
Securing funding for such efforts, the pair warned, is hard. Many chief financial officers are underwhelmed by digital transformation, so building a case for big AI investments will be hard.
Chandrasekaran, in a session titled “Beyond the ChatGPT Hype: Deploying Generative AI in the Enterprise,” rated developing complex AI in-house as the most complex and expensive way to adopt the tech. He suggested that years of work is necessary to blend public and private models, not to mention adding an org’s own data.
But that doesn’t mean it isn’t going to happen. He’s aware of upstart vendors trying to commoditize the tools required to develop and deploy AI in myriad ways – and said venture capitalists are very interested in funding such organizations. ®
